COLUMN: @Play: The Berlin Interpretation
December 18, 2009 12:00 AM |
['@ Play' is a monthly column by John Harris which discusses the history, present and future of the Roguelike dungeon exploring genre. This time - a look at definiting Roguelikes through 'The Berlin Interpretation'.]
Last time when covering Dungeon Hack, I noted that it doesn’t quite fit up to all of the most common definition of a roguelike. While it has random dungeons, hack-and-slash gameplay, and even items that must be identified, it is a first-person game.
And not even an Ultima Underworld kind of first-personness, but the same kind of discrete, right-angled rotation, corridor-centered perspective and step-based movement used in the Wizardry games, which were many years old by that point. And it was a real-time game, too!
For me, the game is obviously rougelike enough to be covered here, since we’re more concerned with what it is that makes roguelikes fun to play than adherance to a laundry list of similarities. But for those who are interested in such classification, we have the Berlin Interpretation.
Arrived at last year at the International Roguelike Development Conference, starting from a document over at Temple of the Roguelike, the Berlin Interpretation is a set of feature descriptions that fairly well encapsulates what a lot of people consider when they think of roguelikes. It covers both graphical and gameplay elements, and has the added advantage of not being posed as a mere checklist. They recognize that some games that are probably roguelike do not meet the exact description presented by the list, and so it is divided into High and Low value factors.
We’re going to take the game through several unusual cases we’ve covered in the past: ToeJam & Earl, Shiren the Wanderer (SNES version) and Dungeon Hack. We’ll also compare Nethack, Dungeon Crawl and Diablo to the list as controls. Let’s have a look!
The original text of the Berlin Interpretation can be found at RogueBasin.
To fold together how roguelike each of these games is, we rate them on a scale from 1 to 5. At the end we add the scores together can compare them to each other. Please note that this system is essentially arbitrary and probably counter to the intended use of the system. I’m using it just to give us a value to compare. This methodology probably wouldn’t stand close observation. For example, I myself have a problem with ASCII graphics being given any kind of priority. So there you go.
High value factors
Random environment generation
The game world is randomly generated in a way that increases replayability. Appearance and placement of items is random. Appearance of monsters is fixed, their placement is random. Fixed content (plots or puzzles or vaults) removes randomness.
Dungeon Hack: 5, levels are surprisingly good, though a little same-ish
TJ&E: 4, levels are fairly complicated but more same-ish until later on
Shiren: 4, levels have more pizazz but some levels are actually static, or drawn from a pool of possibilities
Nethack: 4, its level generator is aging a bit. The “fixed content” thing works against Nethack, which has lots of that.
Dungeon Crawl: 5, the best generator of those presented here
Diablo: 3, good generator visually, but less varied than the others due to a comparative lack of gameplay-relevant dungeon features
Permadeath
You are not expected to win the game with your first character. You start over from the first level when you die. (It is possible to save games but the savefile is deleted upon loading.) The random environment makes this enjoyable rather than punishing.
Dungeon Hack: 2, offers permadeath as a custom option, but unless it’s on for everything it doesn’t make much difference
TJ&E: 4, has a system of lives, but the game is hard enough that many are lost at higher levels and there are no continues, so it works out the same
Shiren: 4, the between-trip continuity options work slightly against it
Nethack: 5, good ol’ permadeath. Note that Nethack contains Discover Mode, which lets players revive after death endlessly, but a Discover win doesn't count as a win to either the community or the high score list.
Dungeon Crawl: 5
Diablo: 2, Diablo 2 introduced permadeath in the form of Hardcore Mode
Turn-based
Each command corresponds to a single action/movement. The game is not sensitive to time, you can take your time to choose your action.
Dungeon Hack: 2, game is a mix of turn-based and real time. Real time wins, generally.
TJ&E: 1, not turn-based at all
Shiren: 5
Nethack: 5
Dungeon Crawl: 5
Diablo: 1
Grid-based
The world is represented by a uniform grid of tiles. Monsters (and the player) take up one tile, regardless of size.
Dungeon Hack: 4, at first the game doesn’t look it, but really it’s the same as most roguelikes, just first person instead of overhead view
TJ&E: 1
Shiren: 5
Nethack: 5
Dungeon Crawl: 5
Diablo: 1
Non-modal
Movement, battle and other actions take place in the same mode. Every action should be available at any point of the game. Violations to this are ADOM's overworld or Angand's and Crawl's shops.
Dungeon Hack: 5
TJ&E: 4, due to mail order
Shiren: 5
Nethack: 5
Dungeon Crawl: 4, due to shops
Diablo: 4, due to shops
(It seems a unfair to punish games for not including a signature feature of Hack, but it's in the description.)
Complexity
The game has enough complexity to allow several solutions to common goals. This is obtained by providing enough item/monster and item/item interactions and is strongly connected to having just one mode.
Dungeon Hack: 2, each class typically has only one solution to a given kind of problem, but often different classes have their own solution. This is against the spirit of the document though.
TJ&E: 4, a lesser variety of solution than other games, but still offers many ways through different situations depending on presents on-hand
Shiren: 5
Nethack 5, and more, it is Nethack’s great strength
Dungeon Crawl: 4, solutions are less universal than in other games
Diablo: 1, most solutions come down to killing things with either swords or spells
Resource management
You have to manage your limited resources (e.g. food, healing potions) and find uses for the resources you receive.
Dungeon Hack: 3, most resources have only one use, but it does make the player rely on them
TJ&E: 4
Shiren: 5
Nethack: 4, Nethack doesn’t hold the player’s feet to the fire as much as predecessors Hack and Rogue
Dungeon Crawl: 5
Diablo: 1, its shops break scarcity
Hack'n'slash
Even though there can be much more to the game, killing lots of monsters is a very important part of a roguelike. The game is player-vs-world: there are no monster/monster relations (like enmities, or diplomacy).
All except TJ&E: 5
TJ&E: 1, the players can only fight monsters using certain presents, and further receive no experience for it
Exploration and discovery
The game requires careful exploration of the dungeon levels and discovery of the usage of unidentified items. This has to be done anew every time the player starts a new game.
Dungeon Hack: 5, surprisingly, the dungeon has a lot of character, there are several important features generated, and the items are much like in Rogue
TJ&E: 4, held back a little from presents being the only kind of item
Shiren: 5
Nethack: 5
Dungeon Crawl: 5
Diablo: 2, item ID is faked. Dungeons have some important random features scattered about though.
Low value factors
Single player character
The player controls a single character. The game is player-centric, the world is viewed through that one character and that character's death is the end of the game.
Dungeon Hack: 5
TJ&E: 5 in single-player, 4 in two-player mode, the game’s two-player mode is a big part of its appeal actually, but it does fall outside the realm of the document. Even in two-player mode the game is still very roguelike-ish, just with a second player played by a second person. Still, many games are played in one-player mode.
Shiren: 5
Nethack: 5
Dungeon Crawl: 5
Diablo: 5 in single-player, 2 outside of it. It beefs up of monsters in multiplayer games necessitating teamwork, which might be good game design but isn’t roguelike so much. Diablo is so strongly focused on multiplayer that I’m rating it as 3 in the aggregate scores.
Monsters are similar to players
Rules that apply to the player apply to monsters as well. They have inventories, equipment, use items, cast spells etc.
Dungeon Hack: 2, no inventories, limited abilities
TJ&E: 1, enemies are very limited compared to players
Shiren: 3, monsters have no inventory, but move much like player
Nethack: 5, intelligent monsters are extremely flexible
Dungeon Crawl: 4, monsters use weapons and armor but only a small number of magic items
Diablo: 1, monsters are simply enemies
Tactical challenge
You have to learn about the tactics before you can make any significant progress. This process repeats itself, i.e. early game knowledge is not enough to beat the late game. (Due to random environments and permanent death, roguelikes are challenging to new players.) The game's focus is on providing tactical challenges (as opposed to strategically working on the big picture, or solving puzzles).
Dungeon Hack: 1, tactically light
TJ&E: 3, getting around opponents through evasion is challenging and fun, and evading danger is a big aspect of any good roguelike. There is little actual combat though.
Shiren: 5, arguably the game’s strongest feature, there are many dangerous situations that can only be escaped through clever use of the tools at hand
Nethack: 4, after the mid-game, many characters can bulldoze through most situations
Dungeon Crawl: 5, tactical combat is a focus of the game
Diablo: 3, focuses on the bad parts of hack-and-slash but still pretty good
ASCII display
The traditional display for roguelikes is to represent the tiled world by ASCII characters.
(I don’t place a lot of stock in this one myself.)
Dungeon Hack: 1
TJ&E: 1
Shiren: 1
Nethack: 5, has an ASCII mode available
Dungeon Crawl: 5
Diablo: 1
Dungeons
Roguelikes contain dungeons, such as levels composed of rooms and corridors.
Dungeon Hack: 5, actually has excellent dungeons, its level builder is among the best
TJ&E: 1, its weird land-and-space levels may be dungeon like, but they aren’t dungeons
Shiren: 3, many of its early areas are outdoors. If that sounds petty, well, it is, but it’s a rather petty criteria.
Nethack: 5
Dungeon Crawl: 5
Diablo: 5, Diablo 2 introduced non-dungeon play environments
Numbers
The numbers used to describe the character (hit points, attributes etc.) are deliberately shown.
Dungeon Hack: 5, HP bar can be switched to numbers and the PC, as a D&D 2E character, flaunts its statistics
TJ&E: 1, only score is visible numerically
Shiren: 5
Nethack: 5
Dungeon Crawl: 5
Diablo: 5, it places a strong emphasis on increasing your stats however you can, usually through equipment
Totals, counting low-value factors for half:
Nethack & Dungeon Crawl (tie): 57.5
Shiren Super Famicom: 54
Dungeon Hack: 42.5
TJ&E: 33
Diablo: 29
The maximum score is 60, the minimum is 12. These scores seem to me to be fairly consistent with the rogue-likeness of each of these games. Most games that, subjectively, have nothing to do with roguelikes would probably score in the low 20s at most, with lots of them coming in at 12. Rogue itself would probably get a 56 or 57, the only thing it lacking being monster similarity to the player (monsters there don’t have inventories).
It seems to objectively be a good measure, but I wonder. Is it possible to devise a game that seems roguelike subjectively, but purposely fails most of the tests of the Berlin Interpretation? Spelunky might be a good game to hold up against it as an edge case. We may come back to this later....
EDIT: Fixed an unfortunately typo.
Categories: Column: At Play
20 Comments
By my scoring, Civ comes out in the mid 30s, X-Com in the low 40s.
Robert | December 18, 2009 4:41 AM
I'd also like to see Demon's Souls for PS3 run through that criteria, by my count it has 30:
-Random environment generation 1
-Permadeath 2, death is costly and you have to die to learn
-Turn-based 1
-Grid-based 1
-Non-modal 5, it lets you die from falling and kill necessary NPCs in 'safe' zones
-Complexity 3, you can find specific items/weapons, learn spells, or just push an enemy off a cliff
-Resource management 1
-Hack'n'slash 5
-Exploration and Discovery 2, exploration is necessary and dangerous, but non-randomized
-Single Player Character 3 (1.5), the multiplayer may provide allies or enemies, depending on your luck
-Monsters are similar to players 3 (1.5), they typically have similar armor, weapon fighting style, and use healing items, but no true inventory
-Tactical challenge 5 (2.5), some of the bosses are brilliant, able to be done very early if you figure out the design
-ASCII display 1 (.5)
-Dungeons 3 (1.5), it's 'outdoor' environments are just rooms without ceilings
-Numbers 5 (2.5)
That puts it above Diablo, but below TJ&E, which sounds about right. It has been compared to rogue on very many occasions (watch the CoOp show), but doesn't feel like a roguelike. It does have heavily roguish elements, like an incredibly dangerous world where tactics and inventory are vastly more important than your character level. Add in more important items and a randomized layout and it could be a contender.
Honestly, I think the scale is good for rogueish games. I think the breakdown would come from using it on patently unroguelike games like Borderlands. Here's to a second round!
Marcus | December 18, 2009 4:51 AM
Great job lining these out! An interesting thought experiment, indeed. Now when people ask me in depth about the definition of and examples for roguelikes, I have somewhere to point them.
MPF | December 18, 2009 7:08 AM
Great article, I enjoy them all.
However, why hate on Diablo? I know Diablo is not that complex but... a 1? Resource management... 1? When I played hardcore I certainly had to manage my potions.. so I'd give that at the very least a 2. Monsters in Diablo.. another 1.. yet Shiren's monsters get a 3? Just seem like decidedly low marks.
Eric | December 18, 2009 8:12 AM
I forgot to add the worst of all being Permadeath. Hardcore mode is now available when you first boot Diablo 2
Eric | December 18, 2009 8:14 AM
Diablo is cleary grid based
alex | December 18, 2009 9:28 AM
I have the same issues with The Berlin Interpretation as I did with the arguments that spawned it, and with some attempts to do the same with versus fighting games.
The original arguments were those of exclusion, not inclusion. The starting seed wasn't things like "Is Diablo a Roguelike?" It was people vehemently arguing that things like Diablo weren't Roguelike.
The accepted games are extremely varied, making defined exclusion/inclusion elements vague, arbitrary, and likely pointless. This is exacerbated by the desire to make a really large list. (And why do you need a large list? So that you can exclude the games you don't want without excluding any accepted games.)
For individual elements, unless they are strict binary yes/no situations, real games tend to fall into hard-to-define areas that will still be contested. (Ask five different people to rate your above list of games and you will get five different numeric results.)
And even some of the "accepted" conventions have changed over time. Once, being ASCII was a major element, even though it didn't directly affect gameplay. It has fallen because of the gradual embracing of graphics. If the next version of Nethack launched without ASCII at all, people wouldn't suddenly question whether it was a Roguelike, which they would have ten years ago.
Baines | December 18, 2009 10:03 AM
Well, any attempt at putting a solid definition on a whole genre is going to fail ultimately. I imagine many coffeebreaks and 7DRLs would especially stretch these criteria to their limits. The Berlin Interpretation is still an interesting start in thinking about the roguelikeness of certain games though.
My own game, Gruesome, scores 38 by my count, which is probably to be expected from a coffeebreak that falls short (deliberately) on many of the complexity rankings.
Darren Grey | December 21, 2009 5:02 AM
I'll give it a shot. Here's my attempt to rank Spelunky on this scale.
Major factors:
Random environment generation 4 - Spelunky's environments are entirely random, but the generator is actually piecing together a grid of pre-defined rooms.
Permadeath 4 - You can buy shortcuts over the course of many plays.
Turn-based 1 - It's a platformer. And a ghost will eat you if you take too long.
Grid-based 2 - The game is blatantly made of tiles, but movement is not constrained to them.
Non-modal 5 - It even has Hack style shops.
Complexity 4 - You can whip or jump on most enemies, but throwing a rock, throwing a bomb, or tricking it to run into a trap are much more effective options.
Resource Management 5 - You can only carry one item.
Hack'n'slash 5
Exploration and discovery 3 - There's no identification. The very random levels with prominent unconnected areas help a lot though.
Single player character 5
Monsters are similar to players 4 - Shopkeepers are pretty scary.
Tactical challenge 4
ASCII display 1
Dungeons 5
Numbers 2 - Just the amount of damage being done.
This gives a total of 43.5. Not bad for a platformer!
Tolmar | December 23, 2009 4:26 PM
So (to offer a perhaps overly arbitrary evaluation) where would we draw the line at inclusion? That is to say, what's the baseline score for a game to be deemed "sufficiently" roguelike?
A quick-and-dirty method could be to say a "true" roguelike scores at least 3 on all major points (and disregards minor points as nonessential), which would set the bar at 30 -- meaning TJ&E qualifies and Diablo falls just short. I'm willing to live with that.
I think a fairer assessment would allow a game to score a 1 on, say, 2-3 major criteria due to deliberate design decisions, lowering the bar to 24-26 -- but that seems a little too low to me.
Of course, this is all completely arbitrary, and probably just a way to justify my own views. I'm definitely of the camp that believes Diablo is not a roguelike, despite cosmetic similarities. But then, I also definitely consider Spelunky to be one, despite being a platformer. (For me, it comes down to the item system.)
I have to admit, I like checklists like this. They make me wonder what tweaks the genre could bear: roguelike but not turn-based (like Spelunky), or roguelike but not hacky (like TJ&E).
Uncle Ovid | December 25, 2009 9:39 AM
Having nothing better to do, I gave Angband a score of 52.5.
-Random environment generation 5: As random as the original Rogue
-Permadeath 5
-Turn-based 5
-Grid-based 5
-Non-modal 4, can't steal from shops
-Complexity 3, it can get rather complex but it's mostly about killing things, surviving things trying to kill you, running away from things trying to kill you and preparing for the these things
-Resource management 3: As new levels are generated and shops are re-stocked there's no "survival horror" element but on the other hand it's filled with difficult equipment choices and even in the end game it's hard be prepared for everything
-Hack'n'slash 5, and then some
-Exploration and Discovery 3, lots of stuff to identify but sources of identification are readily avalible (unless playing ironman), lots of monster attacks to learn but they're not random
-Single Player Character 5
-Monsters are similar to players 4: they can be scared, stunned, blinded, confused, teleported and so on and they can carry items (but they can't use them on you)
-Tactical challenge 5, tactics are pretty all what Angband is about, tactics and treasure hunting
-ASCII display 5, even though graphical options are avalible it seems to be the custom in the community to play it in ASCII
-Dungeons 5, except for the tiny town level you won't see any light of day unless it's summoned by a spell
-Numbers 5, and then some
Nemesis | December 29, 2009 9:06 AM
I resisted adding more games to the article to keep length down, but also because, as I said at the front, it might be possible to create a game that is subjectively roguelike that fails the test.
There is a spark to roguelike games that I have yet to see put into words effectively, although I have tried. I suspect the spark could apply to games that are not roguelikes, too, but we just haven't seen many of those. I think that is why Spelunky is so interesting to us.
What I am trying to say is, my interests lie more towards finding that spark than superficial measurements of rogue-likeness. This may be why I have never been able to get into Angband, as it seems to have less of the spark than most RLs, and why I still play Nethack, which despite several design excesses definitely has it.
John H. | December 29, 2009 4:56 PM
Then perhaps that is a more personal spark, since there are many who would say the opposite :) Roguelikes can be many things to many people, and everyone finds there own spark in different titles, which is why there are so many types of roguelike out there (and why people keep inventing more).
The danger of the sort of list above (and it's far from the only one) is that it can encourage homogeneity rather than allowing diversity to flourish. However it seems, quite thankfully, that developers treat it more as rules to be broken and experimented with than rigidly adhered to.
Darren Grey | December 29, 2009 6:58 PM
Nice article! There's a bug in the link to roguetemple's roguelike definition though ;)
Slash | January 5, 2010 8:29 PM
Darren Grey: Maybe, maybe not. The thing that roguelike games have that I find really appealing exists in some other games, it's just that roguelikes have the best representation of it. But not ALL roguelikes have it. But some games that people represent as being like, or inspired by, roguelikes, like Demon's Souls and Spelunky, seem to have it as well, which leads me to think it may be a universal quality. That, and the spark is a strong source of appeal for me, and it doesn't seem likely (again, to me) that this could result due to an idiosyncrasy.
Slash: Ah, I'll have a look at it later, thanks.
John H. | January 6, 2010 7:46 PM
An interesting edge case might be Legerdemain. I didn't get that far in it, but it seems like it scores pretty low on (at least) random environment generation (most of the maps I found were fixed), permadeath (inn saves), and non-modality (no fighting or casting in towns). And yet, John, I think it had that spark for you, right? If I'd been trying to guess what the spark was, I'd have said "Every game can be different in interesting ways," but I'm not sure how true that is of Legerdemain.
One design choice of Legerdemain, I'd guess, was that if you get rid of random environment generation you pretty much have to get rid of permadeath. It'd just be mean to make the player redo the exact same opening stages after every death.
(By the way, I've put a link to the Berlin Interpretation in my name -- though most of the text is right here in this column!)
matt w | January 15, 2010 6:54 PM
I think Weird Worlds: Return to Infinite Space would be another interesting edge case. I'd score it as follows:
Random environment: 5
Permadeath: 5
Turn-based: 3
Grid-based: 2
Non-modal: 3
Complexity: 4
Resource management: 4
Hack'n'slash: 3
Exploration and discovery: 2
Single PC: 3
Monsters are players: 5
Tactical challenge: 4
ASCII display: 1
Dungeons: 1
Numbers: 1
Total: 38.5
Dave | January 23, 2010 11:19 AM
I think the problem with defining the 'spark' as , is that I think it quite probably applies to a wide variety of boardgames, which would wind up being odd to consider as roguelike to me.
Although, just to see how badly it does on the list, here's my opinion for Settlers of Catan:
Random environment: 2 (Always the same shape, but can get quite varied due to the two variable thing, no identification)
Permadeath: 1 (Everyone keeps playing until the end of the game)
Turn-based: 5 (Nothing can be done without the player who's turn it is)
Grid-based: 5 (Albeit on the corners of hexagons)
Non-modal: 4 (All trades must involve the player who's current turn it is)
Complexity: 3 (Multiple strategies can lead to victory, but they all have the same structure)
Resource management: 5 (The fundamental essence of the game, complete with punishing you for not using your resources fast enough via the robber, and forcing you to prioritise reliability of getting resources with what resources you want access to)
Hack'n'slash: 1 (None.)
Exploration and discovery: 1 (None)
Single PC: 1 (Everyone controls a group of towns, roads, cities, etc)
Monsters are players: 1 (No Monsters)
Tactical challenge: 2 (An experienced player will almost always win over an inexperienced player, but it's a pretty shallow learning curve, you just have to adapt to the environment of each game)
ASCII display: 1 (...They're cardboard tiles and wooden components)
Dungeons: 1 (...Representing an island)
Numbers: 1 (Just score is presented numerically, everything else is presented in pictoral units of 1)
I make that a score of 30.5. Which probably illustrates why using this list in this way is a really bad idea, though I was more looking at Catan as for why the spark probably isn't 'sufficient differences for each game to feel different,' the starting board differences are quite significant, even if the tactical learning curve for how to use them is quite shallow.
Gizensha | February 8, 2010 6:26 PM
Diablo 1 is entirely grid-based, and I'm nearly positive that every entity takes up one grid space.
Funster | February 18, 2010 7:25 AM
Great Dungeon in the Sky (a flash game, really fun)
Random environment: 2 (the levels are randomly chosen from a pool of possible levels.)
Permadeath: 1
Turn Based: 1
Grid: 1
Non-modal: 3 (~10% of your time is spent on the level and character select screen.)
Complex: 5 (how's 300 characters sound?)
Resource Management: 2 (some characters have single use ability.)
Hack n' Slash: 5 (lots of battle!)
Exploring: 2 (might take a while to scope out certain levels.)
1-player: 5
Monsters are players: 5 (monsters are essentially CPU controlled players.)
Tactics: 2 (the most abilities a character can have is 3)
ASCII: 1 (looks like splunkey)
Dungeons: 5 (name really says it all.)
Numbers: 1
Total: 25.5
Might have given points too generously.
milton | November 17, 2010 9:37 PM